Lung cancer is the major cause of cancer death in the United States and the prognosis for most patients with lung cancer has remained poor. The best prognosis is for patients with Stage I NSCLC, but only 50% survive beyond 5 years. Why some patients who present with early stage I lung cancers do very poorly while others do well may be due the heterogeneity of lung tumors assigned to this classification. Advanced in current technologies, which allow the molecular assessment of a large number of genes from individual tumors, have the potential to provide new insight into this important question. The major focus of the proposed studies, is to examined both the mRNA expression profile using cDNA arrays, and the protein expression profiles using quantitative two-dimensional gel electrophoresis (2D gels) in non-small cell lung cancers to devise a novel means of classifying subsets of these tumors. Data analyses will include correlations with clinical and other molecular genetic data and with molecular profiles of other common epithelial tumors to be studied in this project. A network has been set up consisting of investigators with lung cancer expertise to provide the necessary resources for this project. Specific Aim I is to determine a molecular-based classification of tumors that currently belong to stage I lung adenocarcinomas that is more predictive of clinical behavior for this group of tumors. Specific Aim II is to determine a molecular based classification of tumors that currently belong to stage I squamous cell lung carcinomas that is more predictive of clinical behavior of this group of tumors. A total of 200 stage I lung adenocarcinomas and squamous cell carcinomas will be examined during the 5 year period. Data will be analyzed for patterns of gene and protein expression among groups of tumors and with respect to known mutations, clinical outcome, smoking status and comparisons to other common epithelial tumors of the colon and ovary.